| In recent years,the global communication demand capacity in the era of big data has grown exponentially.In order to meet the rapidly growing demand for high-bandwidth network services,the development of high-speed and large-capacity optical fiber communication technology has become an inevitable trend in information communication.However,the Kerr nonlinear effect in the optical fiber will broaden the spectrum of the optical pulse,accelerate the time-domain pulse width broadening,and have a serious impact on the sensitivity of the system.At the same time,the strong nonlinear effect will lead to deviations in signal sampling and reduce the anti-noise capability of the optical communication system..In addition,in the wavelength division multiplexing（WDM）system,the two nonlinear effects of self-phase modulation（SPM）and cross-phase modulation（XPM）will further change the phase of each channel optical signal,thus severely limiting the coherent receiving system.performance.Therefore,to propose a more effective nonlinear impairment compensation method is the key to realize the performance improvement of high-speed,large-capacity and long-distance optical fiber transmission systems.This paper focuses on the research on fiber nonlinear impairment compensation technology based on machine learning.The main research contents are as follows:1.A non-data-aided less-iterations clustering scheme to compensate the nonlinearity for PDM-16QAM/32QAM/64QAM signals.Because the fiber nonlinearity will lead to the drift of data points in the constellation diagram,the standard points of constellation clusters will also change.The optimal centroid of each cluster can be determined by extracting the data local density.This part of the study is based on extracting the local density parameters of the original data,determines the optimal centroid by iterative averaging,and uses the obtained optimal centroid to classify and judge the original data.The effectiveness of the proposed scheme is verified by polarization division multiplexing（PDM）-32-order quadrature amplitude modulation（QAM）/64QAM simulation transmission system and PDM-16QAM experimental transmission system.The simulation and experimental results show that this scheme can reduce the computational complexity of clustering and achieve better bit error rate（BER）performance at different fiber input powers.The PDM-64QAM transmitting 400km is at the optimal fiber input power.The BER value dropped from1.1190×10-4 to 4.0690×10-5,and the BER value dropped from 6.8556×10-3 to 3.7537×10-3at 2d Bm.Finally,the computational complexity of this scheme is briefly analyzed.The computational complexity is O（N*k*t）,where N represents the number of symbols involved in the operation,k represents the order of the modulation format,and t represents the number of algorithm iterations.2.In this part,we firstly applied label propagation algorithm（LPA）to coherent optical communication system for nonlinear impairment compensation,which could exploit the relationship among constellation points to segment the community with only very small amount of label data to predict the label situation on most test data.The effectiveness of the proposed scheme was verified through PDM-16/32/64QAM simulation system.By analyzing the simulation results,we obtained the best propagation coefficient?,which demonstrated that the proposed scheme can effectively make nonlinear decision on received symbols and compensate signal impairment caused by Kerr nonlinear effect and amplified spontaneous emission（ASE）noise,showing a strong robustness and generalization to variations of the transmission environment under different modulation formats,transmission distances,launch power and channel numbers.Then,the proposed scheme was further studies by PDM-16QAM long-haul transmission experiments.When using the label propagation algorithm,the BER can be reduced from 6.47×10-3 to3.79×10-3 at 2d Bm.And the range of fiber input power is expanded by more than 2.6d B.The results demonstrated that the proposed scheme requires no training process and exhibits optimistic compensation performance to the coherent optical communication system with nonlinear impairments.Finally,we analyze the computational complexity of the nonlinear impairment compensation scheme based on the label propagation algorithm in detail.The computational complexity is related to the order of the modulation format,which is about O（m*N）,where m represents the order of the modulation scheme and N represents the number of symbols involved in the operation.3.In the third part,a nonlinear impairment compensation scheme based on improved neural network is proposed.The scheme compensates for the nonlinear impairment caused by self-phase modulation（SPM）and cross-phase modulation（XPM）in WDM systems,and uses overlap and save technology to change the data structure of the input neural network so that the bandwidth broadening caused by large dispersion can be offset.The mathematical model of coupled channel split step Fourier method is applied to the nonlinear activation function of neural network.We verify this scheme by using a5-channel WDM system 28 GBaud PDM-16QAM/64QAM numerical simulation platform.The results show that the proposed scheme has a good compensation effect on nonlinear impairment in WDM system.For example,after 1200km PDM-16QAM transmission,BER of the third channel is reduced from 0.02298 to 0.01025 at 4d Bm launch power.After400km transmission of PDM-64QAM,on the third channel at 4d Bm launch power,this scheme decreases BER from 9.26?10-2 to 2.37?10-2.And compared with the DBP algorithm with the same performance,the complexity can be reduced by two orders of magnitude. |
PDF Full Text Request